The variability of climate is assessed using the observational fifth generation of the European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis (ERA5) for 1980–2019 and data from 35 Coupled Model Intercomparison Project Phase 6 (CMIP6) models for a suite of variables, including the vertically integrated horizontal moisture flux vector from 10 models. The standard deviation of the seasonal anomalies within the 40-year period is calculated for grid points and for rainfall averaged over Australian regions. The four-season average of standard deviation, denoted VAR, is a measure of overall variability. From ERA5, the average for grid points over Australia (AUS) is 4–11% higher than that from all land in 60°S–60°N (LAN) for each variable, except for flux and 850-hPa eastward wind, which are 31–41% higher. The variability linked to two mode indices, NINO34 and PID (Pacific–Indian Dipole), is quantified by the four-season average of the magnitude of the regression anomaly. For each index, the 35-model average (av35) of the linked variability averaged over AUS and LAN for rainfall, temperature and pressure is typically less than 30% of the net VAR. The statistics for 2040–2079 under the shared socioeconomic pathway with expected radiative forcing 5–8.5 W m−2 (SSP5-8.5) are scaled to provide a change representative of the 2°C global warming level. There is little change overall in VAR, except for moisture variables, with flux increasing typically 10%. From av35, for rainfall AUS increased 6% and LAN 9%, in part owing to an increase in the VAR of the NINO34 index of 14%. The PID index provides a stronger link to the range of model change of the variability of all-Australia rainfall, and hence the uncertainty in projected change.
I. G. Watterson (Fri,) studied this question.